CERN-PH-TH/2012-098

Unificaxion

G. F. Giudicea, R. Rattazzib, A. Strumiac,d

(a) CERN, Theory Division, CH-1211 Geneva 23, Switzerland

(b) Institut de Theorie des Phenomenes Physiques, EPFL, Lausanne, Switzerland

(c) Dipartimento di Fisica dell’Universita di Pisa and INFN, Italy

(d) National Institute of Chemical Physics and Biophysics, Tallinn, Estonia

Abstract

Dark matter, gauge coupling unification, and the strong CP prob-

lem find a common and simple solution (in the absence of natu-

ralness) within axion models. We show that such solution, even

without specifying the details of the model implementation, makes

testable predictions for the experimentally measurable axion pa-

rameters: the axion mass and its coupling to photons.

arX

iv:1

204.

5465

v1 [

hep-

ph]

24

Apr

201

2

1 Introduction

Naturalness is considered the leading reason to believe that new physics beyond the Stan-

dard Model (SM) must exist at the weak scale. As no experimental evidence in favor of new

physics at the weak scale has been observed yet, we explore the possibility of discarding

the criterion of naturalness and following the lead of other arguments. Two interesting

motivations to introduce new physics are Dark Matter (DM) and gauge coupling unifica-

tion. Indeed these two arguments, in the absence of naturalness, have led to the intriguing

hypothesis of Split Supersymmetry [1]. In this paper we will show that the same two ar-

guments, together with the additional request of a solution to the strong CP problem, can

lead to another interesting (and more minimal) hypothesis, which offers an experimentally

testable prediction.

To account for DM we assume the existence of the axion, a particle which finds its

motivation in the solution of the strong CP problem. Unlike the cases of the gauge hier-

archy and the cosmological constant, there seems to be no anthropic explanation of the

vanishingly small value of the QCD θ angle. Seeking a natural solution to the strong CP

problem [2] while giving up naturalness on the other two problems appears therefore a

logical option. A large class of invisible axion models (the so-called KSVZ [3]) make use

of new matter, charged under color and PQ symmetry, with mass at about the same scale

as the axion decay constant fa. It is possible that the new matter modifies the running of

the SM coupling constants in such a way to achieve gauge coupling unification. We will

show that this hypothesis, for which we coined the term unificaxion, leads to a prediction

on the ratio between the axion coupling to photons and the the axion mass. Present and

future axion experiments can test the hypothesis of unificaxion, although no new dynamics

beyond the SM is predicted at the weak scale.

This paper is organized as follows. In section 2 we discuss the requirements for new

heavy fermionic particles to achieve unification of SM gauge couplings. These results are

used in section 3 to predict the axion-photon coupling in unificaxion, as a function of the

axion mass. We also extend our discussion to the case of supersymmetry. Finally our

conclusions are drawn in section 4.

2 Unification with matter at an intermediate scale

Let us consider the existence of new fermions Ψ with common masses MΨ. We consider

fermions in real representations of the SM group (because we are interested in very massive

particles) that can be embedded in SU(5) (because we have in mind a grand unified theory).

We want to explore the restrictions on the possible quantum numbers of Ψ by requiring a

2

-4 -2 0 2 4-4

-2

0

2

4

Db3-Db2

Db 2

-D

b 1

G

V

X

L

D

Q

UE

Q

S

T

-2 -1 0 1 2

0

2

4

6

8

Db3-Db2

Db 2

-D

b 1

SM

GUT scale in

GeV

Inte

rmed

iate

scal

ein

GeV

*

103

106

108

1010

1012

1013

10151016

10171018

Figure 1: Left: The lattice formed by all possible values of the one-loop β-function coef-

ficients generated by any combination of fermions in real representations of the SM group

that can be embedded inside SU(5) multiplets. Thick red dots represent the contributions of

single real representations, and the arrows with names represent the simplest of such cases,

which are found inside the following SU(5) representations (see table 1): fundamental (+

conjugate) 5⊕ 5 = L⊕D, antisymmetric (+ conjugate) 10⊕ 10 = Q⊕U ⊕D, symmetric

(+ conjugate) 15⊕ 15 = Q⊕ T ⊕ S, and adjoint 24 = 1⊕ V ⊕G⊕X. Right: The green

area shows the range of the differences between the beta function coefficients for the gauge

couplings, ∆b3−∆b2 and ∆b2−∆b1, that provide unification at a GUT scale between 1016

and 1018 GeV (red dotted lines) with intermediate scale indicated in blue (dashed lines).

The point marked as ? represents the case of low-energy supersymmetry, and its nearest

dot represents the case of split supersymmetry with new scalars at the unification scale.

successful unification of the SM gauge coupling constants at some scale MGUT in one loop

approximation:1

αGUT

=1

αi(MZ)− bSM

i

2πlnMGUT

MZ

− ∆bi2π

lnMGUT

MΨ

. (1)

Here bSMi = 41/10,−19/6,−7 and ∆bi are the contributions to the β-function coefficients

due to SM particles and to new fermions associated with the axion sector, respectively.

Only the values of the differences ∆bi −∆bj are relevant for unification and for fixing

its scale, while adding a universal ∆bi (equal for any i) increases the value of αGUT. The

arrows in fig. 1a show the values of (∆b3 − ∆b2,∆b2 − ∆b1) corresponding to real SM

3

SU(5) SU(3)⊗ SU(2)⊗U(1) n3 n3 n2 z name ∆b3 ∆b2 ∆b1

5⊕ 5 3 1 1/3 0 1 0 0 D 2/3 0 4/15

5⊕ 5 1 2 1/2 0 0 1 0 L 0 2/3 2/5

10⊕ 10 3 1 −2/3 0 1 0 1 U 2/3 0 16/15

10⊕ 10 1 1 −1 0 0 0 1 E 0 0 4/5

10⊕ 10 3 2 1/6 1 0 1 0 Q 4/3 2 2/15

15⊕ 15 3 2 1/6 = = = = Q = = =

15⊕ 15 1 3 1 0 0 2 0 T 0 8/3 12/5

15⊕ 15 6 1 −2/3 2 0 0 0 S 10/3 0 32/15

24 1 3 0 0 0 2 1 V 0 4/3 0

24 8 1 0 1 1 0 0 G 2 0 0

24 3 2 5/6 0 1 1 0 X 4/3 2 10/3

Table 1: The SU(3)⊗SU(2)⊗U(1) quantum numbers, the Dynkin labels for SU(3) (n3, n3)

and SU(2) (n2), and the index z for chiral irreducible representations of the SM group

contained in the SU(5) representations 5 ⊕ 5, 10 ⊕ 10, 15 ⊕ 15, 24. The entries in the

right-hand side give the contributions to ∆bi from fermions in real representations of the

SM group, equal to the previous representations when they are real (Y = 0), or adding

their conjugates when they are chiral (Y 6= 0).

multiplets contained in SU(5) representations with low dimensions: 5⊕5, 10⊕10, 15⊕15,

24. The quantum numbers of these states are summarized in table 1. Within each complete

SU(5) multiplet, arrows sum to zero.

Next, by summing these arrows with integer non-negative coefficients we generate the

most generic set of points produced by arbitrary combinations of these multiplets. The

dots represent all possible combinations: we see that they form a sparse lattice, where each

point can be produced in many different ways. In particular, arbitrary combinations of the

SM representations contained in 5⊕ 5 and 10⊕ 10 are already enough to span the whole

lattice, and nothing more is obtained by introducing the 24 or the 15⊕ 15.

2.1 Extension to arbitrary representations

Here we show that the lattice points shown in fig. 1a describe the most general case for

fermionic matter and that no new points are added by including any other arbitrary irre-

ducible representation R. In general, the representation R can be described by the SU(3)

Dynkin label (n3, n3), the SU(2) Dynkin label (n2), and hypercharge Y . The Dynkin labels

count the differences between the number of boxes in successive rows of the corresponding

Young tableau. So the indices n3, n3, and n2 are non-negative integers. The contributions

4

of Ψ to the β-functions are

∆b3 =d

36

(n2

3 + n23 + n3n3 + 3n3 + 3n3

)∆b2 =

d

18n2(2 + n2) (2)

∆b1 =2

5dY 2,

where d is the dimensionality of R, given by

d = d2d3 d2 = 1 + n2, d3 = (1 + n3)(1 + n3)

(1 +

n3 + n3

2

). (3)

The condition that R is embedded in an irreducible representation of SU(5) implies a

constraint on the possible values of Y . By projecting the weights of a generic multiplet of

SU(5) into the SM subgroup we find that the hypercharge Y must satisfy

Y =n2

2+n3 − n3

3− z. (4)

Here z is the component of a weight of the SU(5) irreducible representation, in the Dynkin

basis, corresponding to the U(1) generator of the Cartan subalgebra. It can be proven that

z can take all possible values in Z.

As a result, ∆bi can only scan a discrete set of values, which is determined by eq.s (2)–

(4), with n2, n3, n3 ∈ N and z ∈ Z. The points obtained by this procedure are shown in

fig. 1a as red thick circles. It is easy to see that all these points can be generated by taking

appropriate combinations of the representations contained in 5⊕ 5 and 10⊕ 10 of SU(5).

2.2 Unification

In fig. 1b we show the range of β-function coefficients compatible with unification of gauge

couplings obtained by varying the intermediate mass MΨ and the GUT scale MGUT as

indicated. The range of acceptable MGUT is limited from below by the requirement of

proton stability [4]

MGUT >

√αGUT

1/242× 1015 GeV. (5)

An upper bound on MGUT of the order of the Planck mass comes from the request that

unification be described at least in leading-log approximation within quantum field theory.

Indeed the request MGUT ∼< MPl is also motivated by gauge/gravity unification.

The green dot shown in fig. 1b represents the one-loop prediction of low-energy su-

persymmetry. Note that the green dot does not exactly correspond to any of the black

5

104 106 108 1010 1012 10141015

1016

1017

1018

1019

Intermediate scale MY in GeV

MG

UT

inG

eV

Excluded by proton decay

0 10 20 30 401015

1016

1017

1018

1019

104

106

108

1010

1012

1014

1ΑGUT

MG

UT

inG

eV

Inte

rmed

iate

scal

eM

Yin

GeV

Figure 2: Left: The lattice points of fig. 1 shown in terms of the intermediate scale MΨ

and the GUT scale. Right: The same points shown in terms of the unification coupling

and mass. The color code identifies the corresponding value of the intermediate scale MΨ,

as indicated. The thick dots within the green band correspond to the range suggested by

gauge/gravity unification, see eq. (6).

dots because, in addition to the fermionic degrees of freedom of gauginos and higgsinos

(G ⊕ V ⊕H), low-energy supersymmetry introduces also some new scalars in incomplete

GUT multiplets (the second Higgs doublet). Notice also that the green dot sits just outside

the region where 1-loop unification is exactly achieved. This is a well know fact, but the

mismatch is well within the size of plausible threshold corrections.

Fig. 2a shows again the same lattice, this time as a function of the GUT scale and of

the intermediate scale. Of course, the discretization of these values holds up to experi-

mental and theoretical uncertainties. Unknown thresholds present at MGUT and/or MΨ

are expected to be equivalent to changing MGUT and MΨ by a factor of a few. Such un-

known thresholds are presumably comparable to two-loop RGE effects, which we neglect.

Furthermore, unknown non-renormalizable operators can give corrections suppressed by

MGUT/MPl.

Fig. 2b shows the prediction for the unified coupling αGUT. A wide variety of values

for αGUT are possible because, for each solution of gauge coupling unification identified

by the points in the region of fig. 1b, one can construct a tower of solutions by adding

complete SU(5) representations, which do not modify MGUT, but increase αGUT. The

6

heavy fermions αGUT MGUT MΨ E/N

Q 1/38 2× 1015 GeV 1× 106 GeV 5/3

2Q 1/38 2× 1015 GeV 5× 1010 GeV 5/3

3Q 1/38 2× 1015 GeV 2× 1012 GeV 5/3

2Q⊕D 1/36 8× 1015 GeV 6× 109 GeV 22/15

2Q⊕ U 1/34 5× 1015 GeV 2× 108 GeV 28/15

G⊕ 2V 1/38 5× 1015 GeV 2× 108 GeV 4/3

Q⊕G⊕ V 1/35 9× 1016 GeV 8× 107 GeV 16/15

Q⊕D ⊕ L 1/36 2× 1015 GeV 1× 106 GeV 2

Table 2: Models of unificaxion with up to 3 fermion multiplets, intermediate mass between

103 and 1014 GeV, and unification mass satisfying eq. (5). Their predictions for αGUT,

MGUT, MΨ, and E/N are shown.

particle content of the simplest models (containing at most 3 of the fermion representations

listed in table 1) are summarized in table 2. These models are selected by requesting that

103 GeV < MΨ < 1014 GeV and that MGUT satisfies eq. (5). It is interesting that some of

these models predict a value of MGUT close to its lower value and thus predict a rate for

proton decay just beyond the present experimental sensitivity.

Furthermore, naıve gauge/gravity unification in 4 dimensions suggests the extra relation

αGUT = k(MGUT

MPl

)2 (6)

valid up to the model-dependent order-one factor k. This relation is shown as a green

band in fig. 2b, where we consider the range 1 < k < 40 with the upper bound motivated

by the heterotic string computation of [5]. It should be stressed that a value k 1 can

be obtained in type I string theory or in M-theory as a result of a parametrically large

volume of compactification. Therefore all values of MGUT ∼< MPl are in principle compatible

with the unification of all forces, though MGUT on the high end does seem perhaps more

plausible, in that it does not require additional very large (or very small) parameters.

3 Axions

In this paper we are focusing on KSVZ axion models [3] which, in addition to the SM

particles, introduce a complex scalar A coupled to new Dirac fermions Ψ, in a representation

of the gauge group Ψ = ⊕rψr. Such models assume a Peccei-Quinn U(1) global symmetry

Ψ→ eiγ5αΨ, A→ e−2iαA, (7)

7

which forbids the Dirac mass term M ΨΨ but allows for Yukawa couplings∑r

λr(AψrPLψr + A†ψrPRψr

). (8)

The U(1) symmetry is spontaneously broken by the vacuum expectation value 〈A〉 = T 2fa,

where TrT aT b = 12T 2δab for the QCD generators of the fermions Ψ, and fa is the decay

constant of the light axion a =√

2 ImA. We assume that all λr couplings have comparable

size, λr ∼ λΨ. Consequently all fermions acquire roughly the same mass MΨ = λΨ〈A〉,representing the only threshold between the weak and unification scales. Remarkably, as

we shall discuss below, the overall size of λΨ does not affect our main prediction.

Non-observation of axion emission from stars and supernovæ implies that the axion

decay constant must be larger than fa > 109 GeV. Furthermore, requiring that the axion

dark matter density generated by the initial misalignment mechanism [6],

Ωa ≈ 0.15

(fa

1012 GeV

)7/6(a∗fa

)2

, (9)

does not exceed the observed dark matter density ΩDM ≈ 0.23 implies the upper bound

fa < 1012 GeV, under the assumption that the axion vev a∗ in the early universe was of the

order of fa [6]. Values of fa ∼ 1012 GeV are therefore favored by the assumption that DM

is made of axions. However, larger values of fa can be compatible with axion dark matter,

if a∗ is sufficiently smaller than its natural value of order fa. Provided that λΨ = O(1),

the heavy fermions Ψ that are associated with the axion dynamics have masses in the

same range as fa but, for small λΨ, the intermediate scale MΨ could be much less than

109 GeV, without conflict with axion bounds from stellar emission. Therefore, a wide range

of intermediate scales MΨ is compatible with axion DM.

3.1 Axion coupling to photons

The anomalous coupling of axion to photons is defined as

− gaγγ4aFµνFµν (10)

where Fµν ≡ 12εµναβFαβ. DM axion experiments are starting to probe the theoretically

favored region of the axion mass ma and axion coupling to photons gaγγ: the ADMX

experiment [7] obtained the limit

|gaγγ| <7

1016 GeV

ma

µeV

√0.3 GeV/cm3

ρDM

(11)

8

0.0 0.5 1.0 1.5 2.0 2.5 3.00

1

2

3

4

Axion coupling EN

Èg aΓΓ

Èma

in10

-16

GeV

ΜeV

Uncertainty band

mu m

d =0.56

±0.05

mu m

d =0.36

Figure 3: The ratio |gaγγ|/ma as a function of the model-dependent coefficient E/N , taking

into account the uncertainty in the quark mass ratios. The darker band corresponds to

mu/md = 0.56±0.05, while the lighter band extends the uncertainty down to mu/md = 0.36,

as claimed in some analyses [8].

for ma in the range ma = 1.9 − 3.55µeV. In eq. (11) we have shown the dependence on

the local DM density, ρDM ≈ 0.3 GeV/cm3, as unprecisely determined from rotation curves

and halo dynamics. We used the most conservative limit corresponding to completely

virialized axions in the galactic halo. Note that, in case of discovery, the uncertainty due

to the galactic axion velocity distribution can be eliminated by studying the frequency

dependence of the signal. A positive signal would allow to precisely measure ma and

determine gaγγ. The experimental determination of ma and gaγγ can provide a crucial test

of the idea of unificaxion, as we now discuss.

The measurable ratio gaγγ/ma is sensitive to the particle content of the theory:

gaγγma

=αem

2πfπmπ

√(1 +

md

mu

)(1 +mu

md

+mu

ms

)

[E

N− 2

3

(4 +mu/md +mu/ms

1 +mu/md +mu/ms

)]. (12)

Here fπ = 93 MeV and we defined E/N =∑

rQPQq2/∑

rQPQT2 where the sums extend

over all fermions ψr with PQ charges QPQ, electric charges q, and TrT aT b = 12T 2δab for the

QCD generators. As we already explained, we are considering models in which all fermions

have the same PQ charges QPQ. Adopting the quark masses given by chiral perturbation

theory at lowest order, mu/md = 0.56 [8], we get

gaγγ =2.0 (E/N − 1.92)

1016 GeV

ma

µeV. (13)

and fig. 3 shows the band induced by the uncertainty on quark masses. Concerning mu/md,

we recall that second-order effects in mq in chiral perturbation theory lead to an uncertainty

9

on the value of mu/md famously known as the Kaplan-Manohar (KM) ambiguity [8].

Conceivably the KM ambiguity could have made mu/md = 0 compatible with experimental

meson masses, thus disposing of the need for an axion. However that possibility is now

disfavored by lattice simulations. Ref. [9] analyzed the issue in detail also considering the

impact of the KM second-order effect in eq. (12). The uncertainty band shown in fig. 3

reflects the assessment of that study. It is also rather evident from eq. (12) that the leading

source of uncertainty is given by the md/mu term in the square root factor up front, which

is singular as mu → 0. In that respect, the error in ms/mu = Q√m2d/m

2u − 1 has a minor

impact and gives no additional uncertainty because Q = 22.7± 0.8 is precisely known and

large.

3.2 Unificaxion prediction for gaγγ/ma

Coming to unificaxion, we find that the model dependent axion coupling coefficient in

eq. (12) is related to the β-function coefficients as

E

N=

∆b2 + 5∆b1/3

∆b3

. (14)

This result is pivotal for our analysis and so it is important to clarify the hypotheses upon

which it rests. One simplifying characteristic of the class of models we are considering

is, as previously mentioned, the existence of a single intermediate threshold at the energy

scale MΨ. Another condition is the existence of just one additional scalar describing the

dynamics of PQ symmetry breaking and the physics at the intermediate threshold. Be-

sides simplicity, the lack of multiple scalar fields may be justified in the multiverse by the

condition of not exacerbating the naturalness problem. Note that scalars charged under

the SM gauge group would affect gauge coupling unification in a way that is completely

independent of axion couplings. Finally one last simplifying assumption, upon which our

conclusions mostly rely, is that all fermions have the same PQ charge, as postulated in

eq. (7). Algebraically this is the statement that the PQ charge matrix QPQ is proportional

to the identity and, in our normalization, QPQ = 1. Because of that, the anomaly coef-

ficients dQAB that control the effective coupling of the axion to the dual field strengths

coincides with the contribution of fermions to the β function coefficients

dQAB = Tr (QPQTA, TB) = T 2δAB. (15)

From this equation, the result in eq. (14) follows immediately. Note that with just one

scalar A in principle one could accommodate fermions with PQ charge = −1 with Yukawa

given by eq. (8), after the replacement A → A†. That would spoil the proportionality

between contributions to the β function and the axion coupling, weakening our conclusions.

10

0 10 20 30 400.5

1.0

1.5

2.0

2.5

1014

1015

1016

1017

1018

1019

1ΑGUT

Axi

onco

uplin

gE

N

GU

Tsc

ale

MG

UT

inG

eV

104 106 108 1010 1012 10140.5

1.0

1.5

2.0

2.510-610-410-21102

0

10

20

30

40

Intermediate scale MY in GeVA

xion

coup

ling

EN

Axion mass in eV for MY = fa

1Α

GU

T

Figure 4: Left: The prediction of each unified model for αGUT and for E/N , the coefficient

entering the axion coupling. The colors indicate the unified mass MGUT. The thick dots are

the points identified in fig. 2 as suggested by gauge/gravity unification, see eq. (6). Right:

The same points expressed in terms of the intermediate scale MΨ, with colors indicating

the value of αGUT. For guidance, we have also translated the intermediate mass MΨ into

the corresponding value of the axion mass ma, under the assumption MΨ = fa.

This could not occur in a supersymmetric theory where, in the presence of just one scalar,

holomorphy implies QPQ = 1. In a non-supersymmetric theory the assumption QPQ = 1

can be justified by assuming that all fermions Ψ sit in the same multiplet within a more

fundamental description, but we will not try to construct explicit examples. It is also worth

remarking that, under the assumption QPQ = 1, the relation between the axion coupling

and the β-function coefficients is preserved, regardless of the dynamics of the PQ breaking

sector and, in particular, regardless of the number and PQ charges of the SM singlets in

that sector.

Equations (13) and (14) provide the link between unification and axion phenomenol-

ogy, which is the key feature of unificaxion. Gauge coupling unification selects a special

range for ∆bi which, in turn, determines the measurable quantity gaγγ/ma. The prediction

for gaγγ/ma is obtained only by the request of unification, with no need to specify the

particular particle content of the model or their interactions. Fig. 4 shows the correlations

between E/N , MGUT, αGUT, and MΨ. In particular, fig. 4b illustrates how the prediction of

unificaxion for E/N , which is directly related to gaγγ/ma through eq. (13), depends on the

11

intermediate mass MΨ. Under the simplifying assumption λΨ ≈ 1, MΨ can be identified

with fa and translated into a value of the axion mass ma (as shown in fig. 4b). However,

the prediction E/N is independent of λΨ and thus more robust. Future experimental de-

terminations of ma and gaγγ/ma will select a region in this plane, allowing for a test of

unificaxion.

It is interesting that unificaxion allows for solutions with an intermediate mass MΨ

compatible with the favored range of fa, giving 1.0 < E/N < 2.5. Moreover, assuming

that the unified coupling is large, as maybe suggested in simpler models of gauge/gravity

unification, we obtain E/N > 1.6. Assuming that the unification scale is very close to the

Planck scale leads to E/N < 2.1. A peculiar coincidence is that the predicted range of

E/N is centered around the value for which gaγγ suffers a perfect cancellation. This is,

of course, a worrisome result, because the axion would turn out to be literally invisible.

Note also that E/N in unificaxion is always smaller than the value predicted by DFSZ

axion models with fermions in complete SU(5) multiplets [10], which is E/N = 8/3, see

eq. (14). Nevertheless, E/N = 8/3 gives a value of the axion-photon coupling which is

inside the range predicted by unificaxion (as evident from eq. (13) and fig. 3, |gaγγ| turns

out to be the same for E/N = 8/3 and E/N = 1.2). Thus, the DFSZ axion or any axion

model with fermions in complete GUT multiplets are experimentally indistinguishable from

unificaxion, unless one devises an experiment sensitive not only to the axion coupling to

photons, but also to its coupling to hadrons.

The ADMX bound in eq. (11) is not yet strong enough to constrain unificaxion. How-

ever, an improved sensitivity is considered feasible, and furthermore new experimental tech-

niques based on cold molecules could allow to probe also lighter axions with fa ∼MGUT [11].

3.3 The supersymmetric case

It may be of interest to extend to the case of supersymmetry our considerations about the

consequences of unification for the axion coupling to photons. Let us consider low-energy

supersymmetry, with new matter in chiral supermultiplets at an intermediate scale MΨ,

associated with the dynamics of the axion sector. In this case, eq. (14) remains valid

because heavy fermions and scalars contribute to ∆bi in the same proportion:

∆bi = ∆bFi + ∆bSi =3

2∆bFi , (16)

where ∆bF,Si are the contributions to the β-function coefficients of the fermion and scalar

components of the chiral supermultiplets. The only change with respect to the SM analysis

amounts to replacing bSMi with bMSSM

i = 33/5, 1,−3. This makes however an important

12

-4 -2 0 2 4-4

-2

0

2

4

Db3-Db2

Db 2

-D

b 1G

V

X

L

D

Q

UE

Q

TMSSM

GU

Tsc

ale in

GeV

for M

Y=

1012

GeV

1015

1016

1017

1018

Figure 5: The GUT lattice as in fig. 1, but now in the supersymmetric case. The region

that allows unification collapses to a narrow strip because unification is already realized in

low-energy supersymmetry without any extra field.

qualitative difference because, as well known, in low-energy supersymmetry unification is

already achieved, at MGUT ≈ 2× 1016 GeV, without the need for extra heavy multiplets.

As a consequence the most plausible scenario for the axion sector is to contain only

complete SU(5) multiplets. In this case (∆b1 = ∆b2 = ∆b3), eq. (14) gives the well-

defined prediction E/N = 8/3. This value is larger than what expected in unificaxion

but, as mentioned previously, it is indistinguishable through measurements of the effective

axion-photon coupling.

However, it is also possible that new matter at the scale MΨ modifies the gauge coupling

evolution, achieving unification at a scale MGUT different than the usual value of low-energy

supersymmetry, 2× 1016 GeV. A solution of the one-loop renormalization-group equation

shows that this happens when

∆b2 −∆b1 =7

5(∆b3 −∆b2) (17)

and that the new unification scale is given by

∆b3 −∆b2 = 4ln(MGUT/2× 1016 GeV)

ln(MGUT/MΨ), (18)

where we worked in the limit of exact unification for low-energy supersymmetry. This is

illustrated by the results shown in fig. 5, which is the analogous of fig. 1b in the case of

13

supersymmetry. The region of ∆bi compatible with gauge coupling unification essentially

collapses to a line, approximately described by eq. (17).

A simple example of modified unification in low-energy supersymmetry is the addition

of heavy chiral supermultiplets in the adjoint in the SM gauge group (G ⊕ V ), a case

motivated by partial N = 2 supersymmetry. In this case, we find that the intermediate

scale MΨ is related to the unification scale by

MΨ =

(1018 GeV

MGUT

)3

2× 1011 GeV. (19)

We also obtain E/N = 2/3, a value smaller than what expected in unificaxion (and thus

a larger value of |gaγγ|).Another example, which gives a more accurate unification of gauge couplings, is the

addition of chiral supermultiplets in the representation G⊕Q. This gives

MΨ =

(1018 GeV

MGUT

)4× 1014 GeV (20)

and, again, E/N = 2/3.

4 Conclusions

The exploration of the ‘multiverse’ prompted theoretical physicists to revisit their belief in

the naturalness criterion and, in some cases, to replace it with a biased statistical approach.

Although interesting from a theoretical point of view, this new paradigm suffers from a

chronic lack of experimental predictions. For instance, when applied to the hierarchy

problem, it leads to the conclusion that it is perfectly acceptable for the Higgs boson

not to be accompanied by other new particles or new dynamics at LHC energies. This

conclusion appears disheartening and it can be hardly used as evidence for the multiverse.

The multiverse hypothesis is desperately in search for observational tests.

One rare exception of this lack of experimental consequences is offered by Split Su-

persymmetry [1]. In this context, new physics should be present at the weak scale, not

because of naturalness, but because of DM and gauge coupling unification. Discovery of a

long-lived gluino at the LHC would provide crucial confirmation of this hypothesis. The

Higgs searches have already narrowed down the possible range of the energy scale of Split

Supersymmetry [12], stating that the gluino lifetime must be τg < ( TeV/mg)5 10−4 sec-

onds. Future searches at the LHC with 14 TeV will probe the existence of a metastable

gluino up to masses of about 2.5–3 TeV.

In this paper, we suggested an alternative approach — unificaxion — for predicting

new physics without invoking naturalness. The idea of unificaxion is to introduce an

14

invisible axion model with a single complex scalar field, which spontaneously breaks a PQ

abelian symmetry at a scale fa and gives masses to a set of fermions. These fermions

contribute to the QCD θ term and have the appropriate quantum numbers to achieve

unification of the SM gauge couplings. In this way, the strong CP problem, DM, and

unification find a common solution with new physics occurring at an intermediate scale

MΨ. While the axion decay constant fa must be larger than about 1011–1012 GeV to

account for DM, the intermediate scale MΨ of the new fermions could conceivably take

even smaller values. Nonetheless, the scale MΨ is expected to be far from the Fermi mass.

Although unificaxion gives no new physics at the LHC, it makes one experimentally testable

prediction in the context of axion searches. The axion-photon coupling is determined by

the particle content that is responsible for gauge coupling unification. In particular, we

have found that 1.0 < E/N < 2.5, implying that

|gaγγ|ma

< 2× 10−16 GeV−1µeV−1. (21)

This prediction is further narrowed down by the hypothesis that gauge unification is ac-

companied by an approximate gauge-gravity unification, which implies 1.6 < E/N < 2.1

and thus|gaγγ|ma

< 1× 10−16 GeV−1µeV−1. (22)

Present axion DM experiments are already probing an interesting range of parameters

and future improvements can reach the sensitivity to test the prediction of unificaxion.

However, it should be stressed that the discovery of an axion that satisfies eq. (21) would

only provide an indication in favor of unificaxion, but not a definitive confirmation. For

example, any axion model in which the new fermions form complete GUT multiplets pre-

dicts E/N = 8/3, leading to a value of |gaγγ| inside the range of eq. (21). Only a test of

the hadronic coupling of the axion could disentangle the two cases. Moreover, a worrisome

feature of the unificaxion prediction is that the axion-photon coupling gaγγ could be van-

ishingly small due to a fortuitous cancellation between short-distance and long-distance

contributions. If this is the case, unificaxion would not be experimentally testable and it

would remain forever buried in the obscurity of the multiverse.

Acknowledgments This work was supported by ESF grant MTT8 and by SF0690030s09

project. We thank C. Scrucca for discussions. The work of R.R. is partly supported by Swiss

National Science Foundation under grant n. 200020-138131.

15

References

[1] N. Arkani-Hamed and S. Dimopoulos, JHEP 0506 (2005) 073 [hep-th/0405159]; G. F. Giu-

dice and A. Romanino, Nucl. Phys. B 699 (2004) 65 [hep-ph/0406088]; N. Arkani-Hamed,

S. Dimopoulos, G. F. Giudice and A. Romanino, Nucl. Phys. B 709 (2005) 3 [hep-

ph/0409232].

[2] R. D. Peccei and H. R. Quinn, Phys. Rev. D 16 (1977) 1791; R. D. Peccei and H. R. Quinn,

Phys. Rev. Lett. 38 (1977) 1440; S. Weinberg, Phys. Rev. Lett. 40 (1978) 223; F. Wilczek,

Phys. Rev. Lett. 40 (1978) 279.

[3] J. E. Kim, Phys. Rev. Lett. 43 (1979) 103; M. A. Shifman, V. I. Vainstein, V. I. Zakharov,

Nucl. Phys. B166 (1980) 4933.

[4] Super-Kamiokande Collaboration, Phys. Rev. Lett. 102 (2009) 141801 [arXiv:0903.0676].

[5] V. S. Kaplunovsky, Nucl. Phys. B 307 (1988) 145 and erratum ibidem B 382 (1992) 436

[hep-th/9205068].

[6] J. Preskill, M. Wise, F. Wilczek, Phys. Lett. B120 (1983) 127; L. Abbott and P. Sikivie,

Phys. Lett. B120 (1983) 133; M. Dine and W. Fischler, Phys. Lett. B120 (1983) 137.

[7] ADMX collaboration, arXiv:0910.5914.

[8] S. Weinberg, Trans. N.Y. Acad. Sci. 38, 185 (1977); D. B. Kaplan and A. V. Manohar,

Phys. Rev. Lett. 56 (1986) 2004; A.V. Manohar and C.T. Sachrajda in the Particle Data

Group.

[9] M.R. Buckley and H. Murayama, JCAP 0707 (2007) 012 [arXiv:0705.0542].

[10] M. Dine, W. Fischler and M. Srednicki, Phys. Lett. B104 (1981) 199; A. P. Zhitnitskii, Sov.

J. Nucl. Phys. 31 (1980) 260.

[11] P. W. Graham and S. Rajendran, Phys. Rev. D 84 (2011) 055013 [arXiv:1101.2691].

[12] G. F. Giudice and A. Strumia, Nucl. Phys. B 858 (2012) 63 [arXiv:1108.6077].

16